Revisiting ‘What to Expect’

I got a note over the weekend from reader Eliah M., who had questions concerning the What to Expect Series I ran recently.  His comments fascinated me, because I generally expect people to be surprised by the low expectations I put on Indy draft picks.  Eliah feels that I’m asking too much of them:

I had some observations regarding the What to Expect articles. Essentially, given your projections and the similarly-drafted player data you provide, it seems to me that you’re predicting most of this year’s rookies to outperform their peers. Looking at last year’s series, that’s a departure from the 2010 projections, which roughly matched the mean values for similarly drafted players (though determining the means was messy and imprecise), except in the case of Angerer, who you projected to produce more tackles than average. Further, it does not seem that last year’s class noticeably outperformed either the 50th percentile for similarly drafted players or your projections.  For Castonzo, Ijalana, Nevis, Carter, and Rucker, you have them slotted to outperform 62.5%, ~70%, ~50%, ~70%, and ~78% of similarly drafted players (by my count), for an average of 66%. Maybe this is intentional or maybe you think the unique attributes of the team or player justifies that in each case, but it struck me and seems borderline contradictory in a series whose intention is to ground expectations in the data.

Eliah then went through the last several selections.  Here are his observations:

2010

Hughes

“All we are looking for this season is…3 to 4 sacks and a forced fumble.” (other expectations were listed but not quantifiable, such as a “spark” when playing)

If we agree that Dwight Freeney’s 13.5 sacks his rookie year represent an outlier and should be excluded, 6 defensive ends were sampled with an average of 3.5 sacks their rookie year.

In other words, you predicted Hughes at exactly the 50th percentile.

In reality, of course, Hughes registered 0 sacks and was the lowest performing defensive end in the sample in this area. He underperformed both the 50th percentile and the prediction.

Angerer

“All we are looking for this season is… 20-30 tackles, 1 special play (INT, FF, sack)”

I’m going to compare this prediction to both the Colts Tampa 2 linebackers you listed and all LBs from the article. For the former, only 2 out of 8 players collected 25 or more tackles as a rookie, placing your projection in the 75th percentile of Colts Tampa 2 rookie LBs (15 was the average number of tackles). Similarly, 6 of 24 LBs overall produced 25 or more tackles, again projecting Angerer to out-produce 75% of rookie LBs (20 tackles was the average).

In other words, you projected Angerer to perform at the 75th percentile.

The real Pat Angerer, however, made 88 tackles, the most by any Colts Tampa 2 rookie LB ever and placing him above 96% of the 24 LBs listed. He outperformed both the 50th percentile and your prediction.

Thomas

“What we are looking for is…30 tackles and an interception… That production would place him in the top half of Colts rookie corners since 2002.”

Ok so obviously we can’t evaluate how this actually turned out, but this jives pretty well with what Colts post-2002 cornerbacks have done, averaging 25 tackles and about an interception.

The prediction matches the 50th percentile (roughly, it’s slightly above). Moving on…

McClendon

“For McClendon, expect…For him to be active for double digit games” and “2-3 starts”

Amongst the 14 Colts interior linemen listed, the average number of games was 9.4 with 3.9 starts, meaning you expected him to outperform slightly in games and underperform barely in starts. The numbers for the rest of the league were so far off the Colts’ that they’re probably not worth evaluating.

I’ll call it a prediction at the mean.

In reality, though, Jacques did not achieve the average number of games and starts, with 5 and 0, respectively. He underperformed both the 50th percentile and the prediction.

Eldridge

The projection here: “fewer than 10 touches.”

The 7 Colts TEs listed averaged 12 receptions their first year. The 4 Colts FBs averaged less than 3. TEs drafted near pick 162 since 2005 grabbed less than 7 balls as rookies. What percentile does less than 10 place Eldridge at? Given the uncertainty of his role and that range of values, unclear.

Mostly for the hell of it, we’ll act like a 50th percentile exists and say the projection matches it.

In reality, Eldridge caught 5 passes. He matched your prediction, with no mean value to compare to.

Mathews

“This is a good pick if…he plays in 6-8 games” and “he posts 10 tackles”

The 9 Colts DTs listed averaged 10 games and 13 tackles. The 10 7th rounders played in 4 games and made 3 tackles. Meaning you expected him to be slightly below the production of the average Colts DT rookie and better than that of the average 7th round DT. Compromise.

Inexact science this is, I’ll say that projection matches the 50th percentile.

Last year, Mathews registered 1 tackle and 8 games. He underperformed both the 50th percentile and your prediction in tackles, but matched each in games played.

Conner

“Expect…10 tackles”

The 9 Colts LBs listed averaged 14 tackles in their rookie season (it’s a slightly different list than the one used for Angerer). The 7th round LBs averaged 6. And the average of 14 and 6 is 10. I like it.

Again, we’ll call it the mean. A little low for a Colts LB, a little high for a 7th rounder.

Like Pat before him, Conner excelled with 36 tackles, more than any of the 7th round LBs and tied for the most amongst Colts players. He is the standard. He outperformed both the 50th percentile and the prediction.

Looking back on 2010, the only projection I think was probably too high was McClendon’s.  Part of what goes into these projections is not only creating a statistical baseline, but also to understand the context into which the player is coming.  McClendon was always seen as a massive project, so I probably should have dialed back expectations for him.  Over all, by Eliah’s reckoning, the Colts had two under-performers (Hughes and McClendon), two over performers (Angerer and Conner), two average performers (Matthews and Eldridge), and one guy who blew out his knee before the season (Thomas).  Over all, I’d say that “What to Expect” was a success in the first year.  I think Eliah agrees, but now his complaint is with my second year projections.

2011

Castonzo

“It’s perfectly reasonable to expect Anthony Castonzo to start every game his rookie year. In fact, given the state of the Colts’ line, anything less would be a legitimate disappointment.”

But only three of the 8 mid to late first round tackles you looked at did that. In other words, you expect Castonzo to outperform 62.5% of mid to late first round offensive tackles in terms of starts.

Simply put, this is nit-picking.  In a sample size so small, 3 of 8 is roughly half.  In addition, I had to take into account two other factors. The first is the draft consensus that Castonzo fell well below where he should have. In other words, it’s reasonable to expect hiim to be on the high end of the projections.  Secondly, the Colts have basically announced he’s getting the starting job.  Given their public words and expectations, it’s reasonable for fans to expect a 16 game starter.  The tiny difference between 62.5% and 50% over just 8 players is vastly outweighed by the context.

Ijalana

“A fair expectation for his rookie year is 16 games started at guard.”

You didn’t give games started numbers for comparable players in this article, but I’m going to assume they’re slightly worse than those for the 8 tackles in the Castonzo post. It’s a guess, but let’s say you’re predicting him to start more than 70% of second round offensive linemen do.

In this case, Eliah didn’t read what I wrote carefully enough.  “Second round tackles have a higher rate of becoming complete busts (there were several far worse than than Tony Ugoh), but more than half of them become starters immediately and hold the job for several years.”  Eliah’s assumption that my projection for Ijalana would be in the 70th percentile is incorrect. The difference is that, as the Colts indicated, he can play at guard or tackle. This is a fairly common practice for second round tackles, and given the problems on the Indy line, it’s perfectly reasonable for him to join that upper half of second round prospects who start immediately somewhere on the line.  The key here is for fans not to worry if he doesn’t win the right tackle job, but to understand that he may be a guard early in his career.  I consider this a 50% projection.

Nevis

“A fair expectation would be for him to make the team and manage to become a rotation player by the end of the year… if he manages to show up in the stat sheet a couple of times and makes even one good play all year (a sack, a tackle for a loss), then he’s had a good rookie year for where he was selected.”

This fits the stats you cited for third round DTs. It looks like you’re projecting him near the 50 percentile.

Good, we agree.

Carter

“It’s reasonable to expect Carter to play in all 16 games and pick up 75-100 carries, with around 400 yards and 3 or four touchdowns. It’s fair to have slightly higher than normal expectations for Carter because of the situation he’s entering.”

Yet 13 of 22 4th round backs considered didn’t even get half that many yards and only 4 made it to 500 yards. How many had 400 yards? I dunno, probably 6 or 7. This is the equivalent of saying Carter should be expected to rush for more yards than (an estimated) 70% of 4th round backs.

(A quick counter to the Kenton Keith example can be made: Donald Brown, first round pick, 78 carries as a rookie, 281 yards. Which is only to say that unique circumstances and individual examples are complicated and sometimes contradictory.)

This is one where Eliah certainly has a point.  I agonized over how to project Carter.  Ultimately, based on rumors I’ve been hearing, Carter may wind up getting more carries than anyone expects.  He’s been given a very defined role in the Indy offense, which is highly unusual for a young back out of the fourth round.  Essentially, I’m playing a combination of a hunch and some inside knowledge to say that slightly higher expectations for Carter are warranted.  A more conservative choice of 25 carries (less than two a game) and 200 yards would have been more in line with average performance for his draft slot, but given what the Colts are saying publicly and privately, that projection simply doesn’t make sense.  This is a guy that they expect to get five carries a game at least. If that happens, he’ll have an easy time approaching the projection.  Note, however, that I said in the piece that given the context the high expectations were fair.

Rucker

“A fair expectation for Rucker is that he makes the team and contributes a dozen tackles over the course of the season. That would be a top 10 season for a 6th round rookie DB since 2004.”

That’s top 10 out of 45, so you expect him to accumulate more tackles than 78% of defensive backs taken in the 6th round.

This is one of those instances where so many DBs taken late flame out, that even the most modest projections look outlandish by comparison.  No one would consider 12 tackles a monster season, but given his context, it would be.  Essentially, when you weed out the complete busts (guys who never made a team), you get about 35% hitting the level I projected for Rucker.  Given Indy’s success rate with DBs (17 of 19 made the team), it’s reasonable to expect Rucker to make the roster at least. Then, if he makes the team we have to project what he’ll do. The Colts system provides a high opportunity for DBs to make tackles. They are looking for guys who tackle well, rather than play ‘shut down’ corner.  Given the fact that he might need to play safety and given the Colts’ obvious needs on special teams, guessing that he might get a dozen or so tackles isn’t outlandish, and might not sound like much, but is actually phenomenal value for his slot.  A more 50% projection would have been for say 5 or 6 tackles, but isn’t that splitting hairs?  The point is that if he does anything at all, it’s a defensible pick.

Ultimately, I essentially graded three of the five picks to have normal performance for their slot, one to vastly out perform it (Carter), and another ‘to vastly outperform’ simply by getting on the field and producing at a very modest level.  I don’t feel that these projections are too high. Eliah’s fear is:

It occurred to me that using starts as a metric could lead to complaining if a player gets injured. Not that there’s a better way to evaluate rookie offensive tackles, it’s just that telling a fan base that has sometimes turned on guys like Sanders, Gonzalez, and Marlin that “anything less (than 16 starts for Castonzo) would be a legitimate disappointment” recommends a caveat, at least to me. 

It’s a valid fear, but I see injuries differently than most people. I don’t blame the player. Obviously, an injured player can’t produce.  I am of the assumption that the vast majority of players want to play and hate being hurt.  Because I don’t think they get hurt on purpose, I don’t hold it against them.  Teams can’t see the future or KNOW who is going to be injured, but for Eliah’s sake, I’ll close this piece by saying:

If a player sustains a significant injury, his performance will be affected (like Kevin Thomas).  Fans should be patient with guys who get hurt.

Quantcast